-- This file:
--   http://anggtwu.net/blogme3/detect-encoding.lua.html
--   http://anggtwu.net/blogme3/detect-encoding.lua
--           (find-angg "blogme3/detect-encoding.lua")
-- Author: Eduardo Ochs <eduardoochs@gmail.com>
--
-- (defun l () (interactive) (find-angg "blogme3/detect-encoding.lua"))

-- «.bad-conversions»	(to "bad-conversions")
-- «.re»		(to "re")



DetectEncoding = Class {
  type = "DetectEncoding",
  new  = function (fname, root)
      local fullfname = (root or "~/")..fname
      local bigstr = ee_readfile(fullfname)
      local lines  = splitlines(bigstr)
      local nlines = #lines
      return DetectEncoding {lines=lines, nlines=nlines, bigstr=bigstr}
    end,
  __index = {
    lastlines0 = function (de, n)
        local b = max(de.nlines - n, 1)
        return table.concat(de.lines, "\n", b, de.nlines)
      end,
    lastlines = function (de, n)
        local ll = de:lastlines0(n or 7)
        if ll:match("\n") then ll = ll:gsub("^(.*\f)([^\f]*)$", "%2") end
        return ll
      end,
    firstandlastlines = function (de, n)
        if de.nlines < 2 then return table.concat(de.lines, "\n") end
        return de.lines[1].."\n"..de:lastlines(n)
      end,
    explicitencoding = function (de, n)
        local str = de:firstandlastlines(n)
        if str:match("utf%-8%-unix")    then return "utf-8-unix" end
        if str:match("raw%-text%-unix") then return "raw-text-unix" end
        if str:match("no%-conversion")  then return "raw-text-unix" end
      end,
    guessencoding0 = function (de)
        local nu8, nother = measure_utf8_ness(de.bigstr)
        de.nu8, de.nother = nu8, nother
        if nu8 == 0 and nother == 0 then return "any" end
        if nu8 >  0 and nother == 0 then return "utf-8-unix" end
        if nu8 == 0 and nother >  0 then return "raw-text-unix" end
      end,
    guessencoding = function (de, n)
        de.enc = de:explicitencoding(n) or de:guessencoding0()
        return de.enc
      end,
  },
}


--[[
* (eepitch-lua51)
* (eepitch-kill)
* (eepitch-lua51)
dofile "detect-encoding.lua"

de = DetectEncoding.new("TODO")
= de:lastlines0(7)
= de:lastlines(7)
= de:lastlines()
= de:firstandlastlines()
= de:explicitencoding()

str = de:firstandlastlines()
= str

= str:match("utf")
= str:match("utf-8")

= DetectEncoding.new("TODO"):guessencoding()

= "\f"

--]]


--[[
* (eepitch-lua51)
* (eepitch-kill)
* (eepitch-lua51)
dofile "detect-encoding.lua"

QUIET = 1
dofile "anggmake.lua"
PPV(anggtranslate)   -- (find-blogme3 "anggmake.lua" "anggtranslate")
PPV(escriptstems)    -- (find-blogme3 "anggmake.lua" "escriptstems")
filelist = map(function (s) return "e/"..s..".e" end, escriptstems)
filelist = anggtranslate
PPV(filelist)

for _,fname in ipairs(filelist) do
  local de = DetectEncoding.new(fname)
  local enc = de:explicitencoding()
  -- print(enc, fname)
  -- if not enc then print(fname) end
  if not enc then
    local nu8, nother = measure_utf8_ness(de.bigstr)
    if nu8 > 0 or nother > 0 then
      -- print(nu8, nother, fname)
      -- if nu8 > 0 and nother >  0 then print(nu8, nother, fname) end
      if nu8 > 0 and nother == 0 then print(nu8, nother, fname) end
    end
  end
end

fa = function (fname) return format('(find-angg "%s")', fname) end

filelist = map(function (s) return "e/"..s..".e" end, escriptstems)
for _,fname in ipairs(filelist) do
  local de = DetectEncoding.new(fname)
  local enc = de:explicitencoding()
  if enc == "raw-text-unix" then print(fa(fname)) end
end

-- (find-angg "LUA/lua50init.lua" "u8c_to_l1")
-- (find-angg "LUA/lua50init.lua" "u8c_to_l1" "measure_utf8_ness")



-- «bad-conversions»  (to ".bad-conversions")
-- Look for bad conversions (like the char c2s here???):
-- (hexl-find-file "~/e/page.e~")
-- (hexl-find-file "~/e/page.e")

* (eepitch-lua51)
* (eepitch-kill)
* (eepitch-lua51)
h = function (s) return tonumber(s, 16) end
h = function (s) return format("%c", tonumber(s, 16)) end
= h"40"

fa = function (fname) return format('(find-fline "%s")', fname) end

QUIET = 1
dofile "anggmake.lua"
-- filelist = map(function (s) return "e/"..s..".e" end, escriptstems)
for _,stem in ipairs(escriptstems) do
  fname = "~/e/"..stem..".e"
  bigstr = ee_readfile(fname)
  if bigstr:match(h"c2") then
    fname2 = "/tmp/"..stem.."-new.e"
    print(fname2)
    bigstr2 = bigstr:gsub(h"c2", "")
    ee_writefile(fname2, bigstr2)
    print(fa(fname))
    print(fa(fname2))
  end
end



end


--]]




-- «re»  (to ".re")
-- (find-es "lua5" "lpeg-quickref")
require "re"


--[[
* (eepitch-lua51)
* (eepitch-kill)
* (eepitch-lua51)
dofile "detect-encoding.lua"

--]]